answerparser.cpp

OSB-PIK-OpenVXI-3.0.0源代码 “中国XML论坛 - 专业的XML技术讨论区--XML在语音技术中的应用”

 /****************License************************************************
  *
  * Copyright 2000-2003.  ScanSoft, Inc.    
  *
  * Use of this software is subject to notices and obligations set forth 
  * in the SpeechWorks Public License - Software Version 1.2 which is 
  * included with this software. 
  *
  * ScanSoft is a registered trademark of ScanSoft, Inc., and OpenSpeech, 
  * SpeechWorks and the SpeechWorks logo are registered trademarks or 
  * trademarks of SpeechWorks International, Inc. in the United States 
  * and other countries.
  *
  ***********************************************************************/
 

 #include "GrammarManager.hpp"
 #include "AnswerParser.hpp"
 #include "CommonExceptions.hpp"
 #include "SimpleLogger.hpp"
 #include "VXIrec.h"                     // for a single type name (eliminate?)
 #include "XMLChConverter.hpp"           // for xmlcharstring
 #include <string>
 #include <sstream>
 #include <map>
 #include <list>
 #include <vector>
 
 #include <framework/MemBufInputSource.hpp>
 #include <sax/ErrorHandler.hpp>         // by XMLErrorReporter
 #include <dom/DOM.hpp>
 #include <sax/SAXParseException.hpp>
 #include <util/XMLChar.hpp>
 using namespace xercesc;
 
 #ifndef WIN32
 inline bool Compare(const XMLCh * a, const XMLCh * b)
 {
   return XMLString::compareString(a, b) == 0;
 }
 #endif
 
 // ---------------------------------------------------------------------------
 class AnswerHolder {
 public:
   AnswerHolder(const XMLCh* ut, const XMLCh* im, 
                const XMLCh* gid, const XMLCh* cf, 
                GrammarManager & gm)
     : utterance(ut), inputmode(im), grammarid(gid), confidence(cf), 
       grammarInfo(0), gramMgr(gm) { }
   
   void SetInstance(DOMElement *is) { instanceDomList.push_back(is); }
   void SetInstance(const xmlcharstring & is){ instanceStrList.push_back(is); }  
 
   bool operator<(const AnswerHolder & x) 
   {
     switch(gramMgr.CompareGrammar(grammarInfo, x.grammarInfo)) {
       case -1: return false;
       default: return true;
     }
   }
 
   bool operator>(const AnswerHolder & x) 
   {
     return !operator<(x);
   }
 
   AnswerHolder & operator=(const AnswerHolder &x)
   {
     if (this != &x) {
       utterance = x.utterance;
       inputmode = x.inputmode;
       grammarid = x.grammarid;
       grammarInfo = x.grammarInfo;
       confidence = x.confidence;
       instanceDomList = x.instanceDomList;
       instanceStrList = x.instanceStrList;
     }
     return *this;
   }
   
   xmlcharstring utterance;
   const XMLCh* inputmode;
   const XMLCh* grammarid;
   const XMLCh* confidence;
   unsigned long grammarInfo;
   GrammarManager & gramMgr;
   
   typedef std::list<DOMElement*> DOMLIST;
   DOMLIST instanceDomList;
   typedef std::list<xmlcharstring> STRLIST;
   STRLIST instanceStrList;    
 };  
 
 typedef std::vector<AnswerHolder > ANSWERHOLDERVECTOR;
 typedef std::map<VXIflt32, ANSWERHOLDERVECTOR, std::greater<VXIflt32> > ANSWERHOLDERMAP;
 
 class AnswerParserErrorReporter : public ErrorHandler {
 public:
   AnswerParserErrorReporter()  { }
   ~AnswerParserErrorReporter() { }
 
   void warning(const SAXParseException& toCatch)     { /* Ignore */ }
   void fatalError(const SAXParseException& toCatch)  { error(toCatch); }
   void resetErrors() { }
 
   void error(const SAXParseException & toCatch)
   { throw SAXParseException(toCatch); }
 
 private:
   AnswerParserErrorReporter(const AnswerParserErrorReporter &);
   void operator=(const AnswerParserErrorReporter &);
 };
 
 // ---------------------------------------------------------------------------
 
 AnswerParser::AnswerParser()
 {
   nlsmlParser = new XercesDOMParser();
   if (nlsmlParser == NULL)
     throw VXIException::OutOfMemory();
 
   nlsmlParser->setValidationScheme( XercesDOMParser::Val_Never);
   nlsmlParser->setDoNamespaces(false);
   nlsmlParser->setDoSchema(false);
   nlsmlParser->setValidationSchemaFullChecking(false);
   nlsmlParser->setCreateEntityReferenceNodes(false);
 
 #pragma message("JC: Look into this")
   // nlsmlParser->setToCreateXMLDeclTypeNode(true);
 
   ErrorHandler *errReporter = new AnswerParserErrorReporter();
   if (errReporter == NULL) {
     delete nlsmlParser;
     throw VXIException::OutOfMemory();
   }
 
   nlsmlParser->setErrorHandler(errReporter);
 }
 
 
 AnswerParser::~AnswerParser()
 {
   if (nlsmlParser != NULL) {
     const ErrorHandler * reporter = nlsmlParser->getErrorHandler();
     nlsmlParser->setErrorHandler(NULL);
     delete reporter;
 
     delete nlsmlParser;
     nlsmlParser = NULL;
   }
 }
 
 inline bool AnswerParser::IsAllWhiteSpace(const XMLCh* str)
 {
   if( !str ) return true;
   while( *str != 0 ) {
     if( *str != '\r' && *str != '\n' && *str != '\t' && *str != ' ')
       return false;
     ++str;  
   }
   return true;
 }
 
 int AnswerParser::Parse(SimpleLogger & log, AnswerTranslator & translator,
                         GrammarManager & gm, int maxnbest,
                         VXIContent * document, vxistring & grammar)
 {
   if (document == NULL) return -1;
 
   // (1) Get document contents.
   const VXIchar * docType;
   const VXIbyte * docContent;
   VXIulong docContentSize = 0;
 
   if (VXIContentValue(document, &docType, &docContent, &docContentSize) !=
       VXIvalue_RESULT_SUCCESS) return -1;
   if (docContent == NULL || docContentSize == 0) return -1;
 
   if (log.IsLogging(3)) {
     VXIString *key = NULL, *value = NULL;
     if (log.LogContent(docType, docContent, docContentSize,
                        &key, &value))
     {
       vxistring temp = VXIStringCStr(key);
       temp += L": ";
       temp += VXIStringCStr(value);
       log.LogDiagnostic(3, temp.c_str());
     }
   }
 
   // (2) Parse NLSML format.
   vxistring type(docType);
 
   if (type == VXIREC_MIMETYPE_XMLRESULT) {
     try {
       VXIcharToXMLCh membufURL(L"nlsml recognition result");
       nlsmlParser->parse(MemBufInputSource(docContent, docContentSize,
                                            membufURL.c_str(), false));
     }
     catch (const XMLException & exception) {
       if (log.IsLogging(0)) {
         log.StartDiagnostic(0) << L"AnswerParser::Parse - XML parsing "
           L"error from DOM: " << XMLChToVXIchar(exception.getMessage());
         log.EndDiagnostic();
       }
       return -3;
     }
     catch (const SAXParseException & exception) {
       if (log.IsLogging(0)) {
         log.StartDiagnostic(0) << L"AnswerParser::Parse - Parse error at line "
                                << exception.getLineNumber()
                                << L", column " << exception.getColumnNumber()
                                << L" - "
                                << XMLChToVXIchar(exception.getMessage());
         log.EndDiagnostic();
       }
       return -3;
     }
 
     switch (ProcessNLSML(log, translator, gm, maxnbest, grammar)) {
     case -1:
       return -3;
     case 0:
       return 0;
     case 1:
       return 1;
     case 2:
       return 2;
     default:
       break;
     }
   }
 
   // (3) Parse other types.
   else {
     return -2;
   }
 
   return 0;
 }
 
 // ---------------------------------------------------------------------------
 // NLSML Parser
 // ---------------------------------------------------------------------------
 
 static const XMLCh NODE_RESULT[]
     = { 'r','e','s','u','l','t','\0' };
 static const XMLCh NODE_INTERPRETATION[]
     = { 'i','n','t','e','r','p','r','e','t','a','t','i','o','n','\0' };
 static const XMLCh NODE_INPUT[]
     = { 'i','n','p','u','t','\0' };
 static const XMLCh NODE_INSTANCE[]
     = { 'i','n','s','t','a','n','c','e','\0' };
 static const XMLCh NODE_NOINPUT[]
     = { 'n','o','i','n','p','u','t','\0' };
 static const XMLCh NODE_NOMATCH[]
     = { 'n','o','m','a','t','c','h','\0' };
 
 static const XMLCh ATTR_GRAMMAR[]
     = { 'g','r','a','m','m','a','r','\0' };
 static const XMLCh ATTR_CONFIDENCE[]
     = { 'c','o','n','f','i','d','e','n','c','e','\0' };
 static const XMLCh ATTR_MODE[]
     = { 'm','o','d','e','\0' };
 
 static const XMLCh DEFAULT_CONFIDENCE[]
     = { '1','0','0','\0' };
 static const XMLCh DEFAULT_MODE[]
     = { 's','p','e','e','c','h','\0' };
 
 
 // This function is used by the other NLSMLSetVars functions to set everything
 // but the 'interpretation'.
 //
 void NLSMLSetSomeVars(vxistring variable, AnswerTranslator & translator,
                       const XMLCh * confidence, const XMLCh *utterance,
                       const XMLCh * mode)
 {
   // (1) Set the 'confidence'.
   vxistring expr = variable + L".confidence = ";
   expr += XMLChToVXIchar(confidence).c_str();
   expr += L" / 100;";
   translator.EvaluateExpression(expr);
 
   // (2) Set the 'utterance'
   translator.SetString(variable + L".utterance",
                        XMLChToVXIchar(utterance).c_str());
 
   // (3) Set the 'inputmode'
   if (Compare(mode, DEFAULT_MODE))
     translator.SetString(variable + L".inputmode", L"voice");
   else
     translator.SetString(variable + L".inputmode",
                          XMLChToVXIchar(mode).c_str());
 }
 
 
 // This sets all of the application.lastresult$ array when the interpretation
 // is a simple result.
 //
 bool NLSMLSetVars(AnswerTranslator & translator, int nbest,
                   const XMLCh * confidence, const XMLCh * utterance,
                   const XMLCh * mode, const XMLCh * interp)
 {
   std::basic_ostringstream<VXIchar> out;
   out << L"application.lastresult$[" << nbest << L"]";
   translator.EvaluateExpression(out.str() + L" = new Object();");
 
   // Set everything but interpretation.
   NLSMLSetSomeVars(out.str(), translator, confidence, utterance, mode);
 
   // Set 'interpretation'.
   translator.SetString(out.str() + L".interpretation",
                        XMLChToVXIchar(interp).c_str());
 
   return true;
 }
 
 // This function recursively processes the data within an <instance>.
 //
 bool NLSMLSetInterp(AnswerTranslator & translator, SimpleLogger & log,
                     vxistring & path, const DOMNode * interp)
 {
   bool foundText = false;
 
   // Look for text first.
   DOMNode * temp = NULL;
   vxistring textValue(L"");
   for (temp = interp->getFirstChild(); temp != NULL;
        temp = temp->getNextSibling())
   {
     if (temp->getNodeType() != DOMNode::TEXT_NODE &&
         temp->getNodeType() != DOMNode::CDATA_SECTION_NODE ) continue;
     foundText = true;
     textValue += XMLChToVXIchar(temp->getNodeValue()).c_str();
   }
   
   if( foundText )
     translator.SetString(path, textValue.c_str());
   
   bool foundElement = false;
 
   // Now try for elements.
   for (temp = interp->getFirstChild(); temp != NULL;
        temp = temp->getNextSibling())
   {
     if (temp->getNodeType() != DOMNode::ELEMENT_NODE) continue;
     if (foundText) {
       log.StartDiagnostic(0)
         << L"AnswerParser::ProcessNLSML - malformed <instance> at "
         << path << L"; mixed text and elements";
       log.EndDiagnostic();
       return false;
     }
     if (!foundElement) {
       translator.EvaluateExpression(path + L" = new Object();");
       foundElement = true;
     }
 
     // Although ECMAScript arrays are treated like properties, calling
     // x.1 = 'val' is invalid.  Instead, x[1] = 'val' must be used.
     XMLChToVXIchar name(temp->getNodeName());
     VXIchar first = *(name.c_str());
     bool isArray = (first=='0' || first=='1' || first=='2' || first=='3' || 
                     first=='4' || first=='5' || first=='6' || first=='7' ||
                     first=='8' || first=='9');
 
     vxistring newPath = path;
     if (isArray)
       newPath += L"[";
     else
       newPath += L".";
     newPath += name.c_str();
     if (isArray)
       newPath += L"]";
 
     NLSMLSetInterp(translator, log, newPath, temp);
   }
 
   if (!foundElement && !foundText) {
     log.StartDiagnostic(0)
       << L"AnswerParser::ProcessNLSML - malformed <instance> at "
       << path << L"; no data found";
     log.EndDiagnostic();
     return false;
   }
 
   return true;
 }